How Nordea’s AI tools help crack the central banks’ code
How Nordea’s AI tools help crack the central banks’ code

How Nordea’s AI tools help crack the central banks’ code

Artificial intelligence and machine learning techniques are changing the game of central bank watching, and Nordea Research is on the front line.

The next time the US Federal Reserve meets to determine its monetary policy moves, it’s not only analysts who will be parsing every word of the meeting minutes. This time the robots will also be listening in – Nordea’s Hawk-o-Meter in particular.

Nordea Research launched the artificial intelligence tool in May to help predict future monetary policy from the Fed, European Central Bank, Bank of England and Sweden’s Riksbank. And the results have been enlightening.

Nordea Markets Chief Analyst Anders Svendsen

Nordea Markets Chief Analyst Anders Svendsen

Close textual analysis of central bank communications is no new phenomenon. All major central banks use communication, from meeting minutes to press conferences, to steer market expectations ahead of future policy action. Investors who can decipher the message faster and more accurately than their competitors have much to gain, but that’s not always easy.

Now Nordea Research’s macro team is putting AI and machine learning to work to help decode the central bankers’ messages.

“We wanted to create a tool that could quickly and objectively convert these central bank texts into data and provide an alternate view to what we, as analysts, do and think normally,” says Anders Svendsen, a chief analyst on the Nordea macro research team and one of the brains behind the self-learning algorithm.

The subject is the theme of the latest Nordea Economic Outlook: “AI-ing central banks”.

How does the Hawk-o-Meter work?

A machine learning algorithm first digests the text of the meeting minutes, distributing it into topics, painting a picture of what issues are top of mind for the bank. The Hawk-o-Meter then takes all words in the text and assigns them a “hawkish” weight, depending on how many times they have been used ahead of rate hikes (hawkish) and rate cuts (dovish) in the past, within a given context.

“There’s no guarantee that the Hawk-o-Meter will show what we as analysts believe. It will be learning on its own, having more words to go on each time,” says Svendsen.

There’s no guarantee that the Hawk-o-Meter will show what we as analysts believe. It will be learning on its own, having more words to go on each time.

Anders Svendsen, Chief Analyst, Nordea Markets

The results of the AI tool have, for the most part, held up. Going back to 2000, the Fed Hawk-o-Meter has correctly predicted 31 rate hikes and 16 rate cuts based on previous minutes, without any wrong signals. However, it failed to predict seven rate hikes and eight rate cuts, including one big miss – the July rate cut. The Fed version has fared slightly better than the Riksbank Hawk-o-Meter, which signalled upcoming rate hikes 20 times since 2005, all correct, and rate cuts 19 times with one false positive, but failed to predict seven hikes and nine cuts.

Results at your fingertips, instantaneously

The AI tool churns through the text of meeting minutes or a press conference immediately after its release, cranking out charts within minutes. One of those is the Hawk-o-Meter, with a scale showing whether central bankers favour a dovish rate cut to stimulate the economy or a hawkish rate cut to slow things down.

A bot publishes the results immediately on @NordeaMacroLab, which you can follow on Twitter or your Bloomberg terminal. You can also subscribe to a slightly more extensive quantitative report, also available in minutes, via Nordea’s e-Markets research portal.

Nordea AI tool predicts monetary policy

Here come the robots

Don’t expect the AI tool to replace human analysts anytime soon. Rather, they add to the narrative analysts are able to tell, says Svendsen.

For example, when the Hawk-o-Meter failed to correctly call the Fed’s July rate cut from its June minutes, analysts had to dig in and understand why. Amid the escalating global trade war, central banks have become more uncertain and more dovish, but it turns out they are not easing for the same reasons they normally do.

Nordea AI tool predicts monetary policy
Anders Svendsen twitter post on Nordea's AI tool that helps predict monetary policy
Nordea Markets Analyst Sofia Fröjd

Nordea Markets Analyst Sofia Fröjd

“All of a sudden, it complements the story and brings value to the whole narrative we’re telling,” Svendsen says. “I think it will be a very strong complement and could lead to efficiency gains down the line if some manual tasks can be transferred to the right digital tool.”

In addition to the Hawk-o-Meter, the analysts also give the minutes a so-called “sentiment score”, where they have chosen the hawkish and dovish words rather than the algorithm. That score tends to be more intuitive than the Hawk-o-Meter, given the analysts’ imprint, according to Sofia Fröjd, a 26-year-old associate in Nordea Research who started working on the project with Nordea as part of her Master’s thesis.

“The sentiment score, unlike the Hawk-o-Meter, will always closely reflect how we as analysts read the minutes,” she says. The sentiment score correctly predicted the July rate cut from the Fed, which the Hawk-o-Meter had failed to see coming.

Nordea AI tool predicts monetary policy

The information provided within this website is intended for background information only. The views and other information provided herein are the current views of Nordea Bank Abp as of the date of publication and are subject to change without notice. The information provided within this website is not an exhaustive description of the described product or the risks related to it, and it should not be relied on as such, nor is it a substitute for the judgement of the recipient.

The information provided within this website is not intended to constitute and does not constitute investment advice nor is the information intended as an offer or solicitation for the purchase or sale of any financial instrument. The information provided within this website has no regard to the specific investment objectives, the financial situation or particular needs of any particular recipient. Relevant and specific professional advice should always be obtained before making any investment or credit decision. It is important to note that past performance is not indicative of future results.

Nordea Bank Abp is not and does not purport to be an adviser as to legal, taxation, accounting or regulatory matters in any jurisdiction.

The information provided within this website may not be reproduced, distributed or published for any purpose without the prior written consent from Nordea Bank Abp.

Related articles